- **TFLOPS:** Stands for "Tera Floating Point Operations Per Second." - **Measure of Performance:** Indicates how many trillion floating-point operations a processor can handle per second. - **Floating-Point Operations:** Calculations on real numbers, crucial in scientific computations and machine learning. - **Hardware Capability:** TFLOPS is a hardware metric, reflecting the computational power of GPUs and CPUs. - **Real-World Performance:** Actual performance in tasks (like AI processing) depends on how efficiently software uses this hardware capability. - **Comparison Metric:** Useful for comparing computational power across different systems and architectures. - **Applications:** Key in fields requiring heavy computations, such as gaming, 3D rendering, and AI model training and inference. #cheatsheet #llm #ai #computing